resolved conflicts

example/02_gemm_bilinear/CMakeLists.txt

-list(APPEND gpu_list1 gfx1100 gfx1101 gfx1102)
-list(APPEND gpu_list2 gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list1 AND target EQUAL 0)
-    add_example_executable(example_gemm_bilinear_wmma_fp16 gemm_bilinear_wmma_fp16.cpp)
-    add_example_executable(example_gemm_bilinear_wmma_int8 gemm_bilinear_wmma_int8.cpp)
-endif()
-if(GPU_TARGETS MATCHES "gfx908" OR GPU_TARGETS MATCHES "gfx90a" OR GPU_TARGETS MATCHES "gfx940")
-   set(target 1)
- endif()
-endforeach()
-
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list2 AND target EQUAL 0)
-    add_example_executable(example_gemm_bilinear_xdl_fp16 gemm_bilinear_xdl_fp16.cpp)
-   set(target 1)
- endif()
-endforeach()
+add_example_executable(example_gemm_bilinear_wmma_fp16 gemm_bilinear_wmma_fp16.cpp)
+add_example_executable(example_gemm_bilinear_wmma_int8 gemm_bilinear_wmma_int8.cpp)
+add_example_executable(example_gemm_bilinear_xdl_fp16 gemm_bilinear_xdl_fp16.cpp)

example/03_gemm_bias_relu/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-    add_example_executable(example_gemm_bias_relu_xdl_fp16 gemm_bias_relu_xdl_fp16.cpp)
-   set(target 1)
- endif()
-endforeach()
+add_example_executable(example_gemm_bias_relu_xdl_fp16 gemm_bias_relu_xdl_fp16.cpp)

example/04_gemm_add_add_fastgelu/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list AND target EQUAL 0)
-        add_custom_target(example_gemm_add_add_fastgelu_xdl)
-        add_example_executable(example_gemm_add_add_fastgelu_xdl_bf16 gemm_add_add_fastgelu_xdl_bf16.cpp)
-        add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_bf16)
+add_custom_target(example_gemm_add_add_fastgelu_xdl)
+add_example_executable(example_gemm_add_add_fastgelu_xdl_bf16 gemm_add_add_fastgelu_xdl_bf16.cpp)
+add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_bf16)
 
-        add_example_executable(example_gemm_add_add_fastgelu_xdl_fp16 gemm_add_add_fastgelu_xdl_fp16.cpp)
-        add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_fp16)
+add_example_executable(example_gemm_add_add_fastgelu_xdl_fp16 gemm_add_add_fastgelu_xdl_fp16.cpp)
+add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_fp16)
 
-        add_example_executable(example_gemm_add_add_fastgelu_xdl_fp32 gemm_add_add_fastgelu_xdl_fp32.cpp)
-        add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_fp32)
+add_example_executable(example_gemm_add_add_fastgelu_xdl_fp32 gemm_add_add_fastgelu_xdl_fp32.cpp)
+add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_fp32)
 
-        if(USE_BITINT_EXTENSION_INT4)
-            add_example_executable(example_gemm_add_add_fastgelu_xdl_int4 gemm_add_add_fastgelu_xdl_int4.cpp)
-            add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_int4)
-        endif(USE_BITINT_EXTENSION_INT4)
-
-        add_example_executable(example_gemm_add_add_fastgelu_xdl_int8 gemm_add_add_fastgelu_xdl_int8.cpp)
-        add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_int8)
-        set(target 1)
-    endif()
-endforeach()
+add_example_executable(example_gemm_add_add_fastgelu_xdl_int8 gemm_add_add_fastgelu_xdl_int8.cpp)
+add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_int8)
 
-set(gpu_list "")
+if(USE_BITINT_EXTENSION_INT4)
+    add_example_executable(example_gemm_add_add_fastgelu_xdl_int4 gemm_add_add_fastgelu_xdl_int4.cpp)
+    add_example_dependencies(example_gemm_add_add_fastgelu_xdl example_gemm_add_add_fastgelu_xdl_int4)
+endif(USE_BITINT_EXTENSION_INT4)
 
 list(APPEND gpu_list gfx90a gfx940 gfx941 gfx942)
 set(target 0)

example/09_convnd_fwd/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-      add_example_executable(example_convnd_fwd_xdl_fp32 convnd_fwd_xdl_fp32.cpp)
-      add_example_executable(example_convnd_fwd_xdl_fp16 convnd_fwd_xdl_fp16.cpp)
-      add_example_executable(example_convnd_fwd_xdl_bf16 convnd_fwd_xdl_bf16.cpp)
-      add_example_executable(example_convnd_fwd_xdl_int8 convnd_fwd_xdl_int8.cpp)
-      add_example_executable(example_convnd_fwd_xdl_fp8 convnd_fwd_xdl_fp8.cpp)
-      add_example_executable(example_convnd_fwd_xdl_bf8 convnd_fwd_xdl_bf8.cpp)
-      # FIXME: re-enable this exampe as test when SWDEV-335738 is fixed
-      add_example_executable_no_testing(example_convnd_fwd_xdl_fp64 convnd_fwd_xdl_fp64.cpp)
-   set(target 1)
- endif()
-endforeach()
-
+add_example_executable(example_convnd_fwd_xdl_fp32 convnd_fwd_xdl_fp32.cpp)
+add_example_executable(example_convnd_fwd_xdl_fp16 convnd_fwd_xdl_fp16.cpp)
+add_example_executable(example_convnd_fwd_xdl_bf16 convnd_fwd_xdl_bf16.cpp)
+add_example_executable(example_convnd_fwd_xdl_int8 convnd_fwd_xdl_int8.cpp)
+add_example_executable(example_convnd_fwd_xdl_fp8 convnd_fwd_xdl_fp8.cpp)
+add_example_executable(example_convnd_fwd_xdl_fp64 convnd_fwd_xdl_fp64.cpp)
+add_example_executable(example_convnd_fwd_xdl_bf8 convnd_fwd_xdl_bf8.cpp)
+add_example_executable(example_convnd_fwd_xdl_fp8_bf8 convnd_fwd_xdl_fp8_bf8.cpp)
 add_example_executable(example_convnd_fwd_dl_fp16 convnd_fwd_dl_fp16.cpp)
 add_example_executable(example_convnd_fwd_dl_fp32 convnd_fwd_dl_fp32.cpp)
 add_example_executable(example_convnd_fwd_dl_int8 convnd_fwd_dl_int8.cpp)

example/09_convnd_fwd/convnd_fwd_xdl_fp8_bf8.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "convnd_fwd_common.hpp"
+
+#include "ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp"
+
+#include "ck/library/utility/convolution_host_tensor_descriptor_helper.hpp"
+
+using InDataType       = ck::f8_t;
+using WeiDataType      = ck::bf8_t;
+using AccDataType      = float;
+using CShuffleDataType = ck::f8_t;
+using OutDataType      = ck::f8_t;
+using AComputeType     = ck::f8_t;
+using BComputeType     = ck::bf8_t;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using InElementOp  = ck::tensor_operation::element_wise::PassThrough;
+using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
+using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvSpec =
+    ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
+
+static constexpr auto GemmSpec = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
+
+template <ck::index_t NDimSpatial, typename InLayout, typename WeiLayout, typename OutLayout>
+using DeviceGroupedConvNDFwdInstance =
+    ck::tensor_operation::device::DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle<
+        NDimSpatial,
+        InLayout,
+        WeiLayout,
+        ck::Tuple<>,
+        OutLayout,
+        InDataType,
+        WeiDataType,
+        AccDataType,
+        CShuffleDataType,
+        ck::Tuple<>,
+        OutDataType,
+        InElementOp,
+        WeiElementOp,
+        OutElementOp,
+        ConvSpec,    // ConvForwardSpecialization
+        GemmSpec,    // GemmSpecialization
+        1,           //
+        256,         // BlockSize
+        128,         // MPerBlock
+        256,         // NPerBlock
+        32,          // KPerBlock
+        8,           // AK1
+        8,           // BK1
+        32,          // MPerXdl
+        32,          // NPerXdl
+        2,           // MXdlPerWave
+        4,           // NXdlPerWave
+        S<4, 64, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
+        S<1, 0, 2>,  // ABlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // ABlockTransferSrcAccessOrder
+        2,           // ABlockTransferSrcVectorDim
+        8,           // ABlockTransferSrcScalarPerVector
+        8,           // ABlockTransferDstScalarPerVector_AK1
+        1,           // ABlockLdsExtraM
+        S<4, 64, 1>, // BBlockTransferThreadClusterLengths_BK0_N_BK1
+        S<1, 0, 2>,  // BBlockTransferThreadClusterArrangeOrder
+        S<1, 0, 2>,  // BBlockTransferSrcAccessOrder
+        2,           // BBlockTransferSrcVectorDim
+        8,           // BBlockTransferSrcScalarPerVector
+        8,           // BBlockTransferDstScalarPerVector_BK1
+        1,           // BBlockLdsExtraN
+        1,
+        1,
+        S<1, 32, 1, 8>,
+        8,
+        AComputeType,
+        BComputeType>;
+
+#include "run_convnd_fwd_example.inc"
+
+int main(int argc, char* argv[]) { return run_convnd_fwd_example(argc, argv) ? 0 : 1; }

example/10_convnd_fwd_multiple_d_multiple_reduce/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list AND target EQUAL 0)
-        add_custom_target(example_convnd_fwd_reduce_xdl)
+add_custom_target(example_convnd_fwd_reduce_xdl)
+add_example_executable(example_convnd_fwd_max_xdl_int8 convnd_fwd_max_xdl_int8.cpp)
+add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_int8)
 
-        add_example_executable(example_convnd_fwd_max_xdl_int8 convnd_fwd_max_xdl_int8.cpp)
-        add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_int8)
+add_example_executable_no_testing(example_convnd_fwd_max_xdl_bf16 convnd_fwd_max_xdl_bf16.cpp)
+add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_bf16)
 
-        add_example_executable_no_testing(example_convnd_fwd_max_xdl_bf16 convnd_fwd_max_xdl_bf16.cpp)
-        add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_bf16)
+add_example_executable_no_testing(example_convnd_fwd_max_xdl_fp16 convnd_fwd_max_xdl_fp16.cpp)
+add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_fp16)
 
-        add_example_executable_no_testing(example_convnd_fwd_max_xdl_fp16 convnd_fwd_max_xdl_fp16.cpp)
-        add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_fp16)
+add_example_executable(example_convnd_fwd_max_xdl_fp32 convnd_fwd_max_xdl_fp32.cpp)
+add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_fp32)
 
-        add_example_executable(example_convnd_fwd_max_xdl_fp32 convnd_fwd_max_xdl_fp32.cpp)
-        add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_fp32)
-
-        if(USE_BITINT_EXTENSION_INT4)
-            add_example_executable(example_convnd_fwd_max_xdl_int4 convnd_fwd_max_xdl_int4.cpp)
-            add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_int4)
-        endif(USE_BITINT_EXTENSION_INT4)
-        set(target 1)
-    endif()
-endforeach()
+if(USE_BITINT_EXTENSION_INT4)
+    add_example_executable(example_convnd_fwd_max_xdl_int4 convnd_fwd_max_xdl_int4.cpp)
+    add_example_dependencies(example_convnd_fwd_reduce_xdl example_convnd_fwd_max_xdl_int4)
+endif(USE_BITINT_EXTENSION_INT4)

example/14_gemm_quantization/CMakeLists.txt

-# dlops
 add_example_executable(example_gemm_dl_quantization_int8 gemm_dl_quantization_int8.cpp)
-# xdlops
-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-   add_example_executable(example_gemm_xdl_bias_relu_quantization_int8 gemm_xdl_bias_relu_quantization_int8.cpp)
-   add_example_executable(example_gemm_xdl_quantization_int8 gemm_xdl_quantization_int8.cpp)
-   set(target 1)
- endif()
-endforeach()
+add_example_executable(example_gemm_xdl_bias_relu_quantization_int8 gemm_xdl_bias_relu_quantization_int8.cpp)
+add_example_executable(example_gemm_xdl_quantization_int8 gemm_xdl_quantization_int8.cpp)

example/15_grouped_gemm/CMakeLists.txt

@@ -23,8 +23,8 @@ add_example_dependencies(example_grouped_gemm_xdl example_grouped_gemm_xdl_bf16)
 add_example_executable(example_grouped_gemm_xdl_int8 grouped_gemm_xdl_int8.cpp)
 add_example_dependencies(example_grouped_gemm_xdl example_grouped_gemm_xdl_int8)
 
-add_example_executable(example_grouped_gemm_xdl_fixed_nk_fp8 grouped_gemm_xdl_fixed_nk_fp8.cpp)
-add_example_dependencies(example_grouped_gemm_xdl example_grouped_gemm_xdl_fixed_nk_fp8)
+add_example_executable(example_grouped_gemm_xdl_fixed_nk_fp16_fp8 grouped_gemm_xdl_fixed_nk_fp16_fp8.cpp)
+add_example_dependencies(example_grouped_gemm_xdl example_grouped_gemm_xdl_fixed_nk_fp16_fp8)
 
 if(USE_BITINT_EXTENSION_INT4)
     add_example_executable(example_grouped_gemm_xdl_int4 grouped_gemm_xdl_int4.cpp)

example/15_grouped_gemm/grouped_gemm_xdl_fixed_nk_fp16.cpp

@@ -36,7 +36,7 @@ using BDataType        = F16;
 using AccDataType      = F32;
 using CShuffleDataType = F32;
 using DsDataType       = ck::Tuple<>;
-using EDataType        = F32;
+using EDataType        = F16;
 
 using ALayout  = Row;
 using BLayout  = Col;
@@ -55,7 +55,7 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemm_Xdl_F
 //######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
 //######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
 //######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,              8,              8,         1,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,             3,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              4>;
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,              8,              8,         1,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,             3,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               4>;
 // clang-format on
 
 struct ProblemSize final
@@ -298,9 +298,9 @@ int main(int argc, char* argv[])
 
     for(int i = 0; i < problem_size.group_count; i++)
     {
-        problem_size.Ms.push_back(256 + 256 * i);
-        problem_size.Ns.push_back(256);
-        problem_size.Ks.push_back(128);
+        problem_size.Ms.push_back(128 + rand() % 128);
+        problem_size.Ns.push_back(1024);
+        problem_size.Ks.push_back(1024);
 
         problem_size.stride_As.push_back(problem_size.Ks[i]);
         problem_size.stride_Bs.push_back(problem_size.Ks[i]);

example/15_grouped_gemm/grouped_gemm_xdl_fixed_nk_fp8.cpp → example/15_grouped_gemm/grouped_gemm_xdl_fixed_nk_fp16_fp8.cpp

@@ -35,7 +35,7 @@ using PassThrough = ck::tensor_operation::element_wise::PassThrough;
 using ADataType        = F16;
 using BDataType        = F8;
 using AccDataType      = F32;
-using CShuffleDataType = F32;
+using CShuffleDataType = F16;
 using DsDataType       = ck::Tuple<>;
 using EDataType        = F16;
 
@@ -56,7 +56,7 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGroupedGemm_Xdl_F
 //######|        |        |         |        |      Type|      Type|        Type|         DataType|       Type|      Type| Elementwise| Elementwise|  Elementwise| Spacialization| Prefetch|  Size| Block| Block| Block|    |    |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|         _MBlock_MWaveMPerXdl| ScalarPerVector|
 //######|        |        |         |        |          |          |            |                 |           |          |   Operation|   Operation|    Operation|               |    Stage|      |      |      |      |    |    |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|         _NBlock_NWaveNPerXdl|   _NWaveNPerXdl|
 //######|        |        |         |        |          |          |            |                 |           |          |            |            |             |               |         |      |      |      |      |    |    |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            |                             |                |
-        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,              8,              8,         1,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,             3,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,              8>;
+        < ALayout, BLayout, DsLayout, ELayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EDataType,  AElementOp,  BElementOp, CDEElementOp,    GemmDefault,        1,   256,    64,   128,    32,   8,   8,   32,   32,    1,    2,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,              3,              8,              8,         1,  S<1, 4, 64, 1>,  S<0, 2, 1, 3>,  S<0, 2, 1, 3>,             3,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
 struct ProblemSize final

example/16_gemm_multi_d_multi_reduces/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list AND target EQUAL 0)
-        add_custom_target(example_gemm_reduce_xdl)
-        add_custom_target(example_gemm_reduce_xdl_max)
-        add_custom_target(example_gemm_reduce_xdl_mean_meansquare)
-        add_custom_target(example_gemm_add_add_mean_meansquare_xdl)
-
-        add_example_executable(example_gemm_max_xdl_fp16 gemm_max_xdl_fp16.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_fp16)
-
-        add_example_executable(example_gemm_add_add_mean_meansquare_xdl_fp16 gemm_add_add_mean_meansquare_xdl_fp16.cpp)
-        add_example_dependencies(example_gemm_add_add_mean_meansquare_xdl example_gemm_add_add_mean_meansquare_xdl_fp16)
-
-        add_example_executable(example_gemm_mean_meansquare_xdl_fp16 gemm_mean_meansquare_xdl_fp16.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_fp16)
-
-        add_example_executable(example_gemm_max_xdl_int8 gemm_max_xdl_int8.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_int8)
-
-        add_example_executable(example_gemm_add_addsquare_xdl_int8 gemm_add_addsquare_xdl_int8.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_add_addsquare_xdl_int8)
-
-        add_example_executable(example_gemm_max_xdl_fp32 gemm_max_xdl_fp32.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_fp32)
-
-        add_example_executable(example_gemm_mean_meansquare_xdl_fp32 gemm_mean_meansquare_xdl_fp32.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_fp32)
-
-        add_example_executable(example_gemm_max_xdl_bf16 gemm_max_xdl_bf16.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_bf16)
-
-        add_example_executable(example_gemm_mean_meansquare_xdl_bf16 gemm_mean_meansquare_xdl_bf16.cpp)
-        add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_bf16)
-
-        add_example_dependencies(example_gemm_reduce_xdl
-            example_gemm_reduce_xdl_mean_meansquare
-            example_gemm_reduce_xdl_max
-            example_gemm_add_add_mean_meansquare_xdl)
-
-        if(USE_BITINT_EXTENSION_INT4)
-            add_example_executable(example_gemm_max_xdl_int4 gemm_max_xdl_int4.cpp)
-            add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_int4)
-        endif()
-        set(target 1)
-    endif()
-endforeach()
+add_custom_target(example_gemm_reduce_xdl)
+add_custom_target(example_gemm_reduce_xdl_max)
+add_custom_target(example_gemm_reduce_xdl_mean_meansquare)
+add_custom_target(example_gemm_add_add_mean_meansquare_xdl)
+
+add_example_executable(example_gemm_max_xdl_fp16 gemm_max_xdl_fp16.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_fp16)
+
+add_example_executable(example_gemm_add_add_mean_meansquare_xdl_fp16 gemm_add_add_mean_meansquare_xdl_fp16.cpp)
+add_example_dependencies(example_gemm_add_add_mean_meansquare_xdl example_gemm_add_add_mean_meansquare_xdl_fp16)
+
+add_example_executable(example_gemm_mean_meansquare_xdl_fp16 gemm_mean_meansquare_xdl_fp16.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_fp16)
+
+add_example_executable(example_gemm_max_xdl_int8 gemm_max_xdl_int8.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_int8)
+
+add_example_executable(example_gemm_add_addsquare_xdl_int8 gemm_add_addsquare_xdl_int8.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_add_addsquare_xdl_int8)
+
+add_example_executable(example_gemm_max_xdl_fp32 gemm_max_xdl_fp32.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_fp32)
+
+add_example_executable(example_gemm_mean_meansquare_xdl_fp32 gemm_mean_meansquare_xdl_fp32.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_fp32)
+
+add_example_executable(example_gemm_max_xdl_bf16 gemm_max_xdl_bf16.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_bf16)
+
+add_example_executable(example_gemm_mean_meansquare_xdl_bf16 gemm_mean_meansquare_xdl_bf16.cpp)
+add_example_dependencies(example_gemm_reduce_xdl_mean_meansquare example_gemm_mean_meansquare_xdl_bf16)
+
+add_example_dependencies(example_gemm_reduce_xdl
+    example_gemm_reduce_xdl_mean_meansquare
+    example_gemm_reduce_xdl_max
+    example_gemm_add_add_mean_meansquare_xdl)
+
+if(USE_BITINT_EXTENSION_INT4)
+    add_example_executable(example_gemm_max_xdl_int4 gemm_max_xdl_int4.cpp)
+    add_example_dependencies(example_gemm_reduce_xdl_max example_gemm_max_xdl_int4)
+endif()

example/17_convnd_bwd_data/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-   add_example_executable(example_convnd_bwd_data_xdl_fp16 convnd_bwd_data_xdl_fp16.cpp)
-   if(result EQUAL 0)
-      target_link_libraries(example_convnd_bwd_data_xdl_fp16 PRIVATE utility)
-   endif()
-   set(target 1)
- endif()
-endforeach()
+add_example_executable(example_convnd_bwd_data_xdl_fp16 convnd_bwd_data_xdl_fp16.cpp)
+if(result EQUAL 0)
+    target_link_libraries(example_convnd_bwd_data_xdl_fp16 PRIVATE utility)
+endif()
 
 add_example_executable(example_convnd_bwd_data_dl_fp16 convnd_bwd_data_dl_fp16.cpp)
 if(result EQUAL 0)

example/20_grouped_conv_bwd_weight/CMakeLists.txt

-list(APPEND gpu_list_xdl gfx908 gfx90a gfx940 gfx941 gfx942)
-list(APPEND gpu_list_wmma gfx1100 gfx1101 gfx1102)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list_xdl AND target EQUAL 0)
-        add_custom_target(example_grouped_conv_bwd_weight)
-        add_example_executable(example_grouped_conv_bwd_weight_xdl_fp16 grouped_conv_bwd_weight_xdl_fp16.cpp)
-        add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_fp16)
+add_custom_target(example_grouped_conv_bwd_weight)
+add_example_executable(example_grouped_conv_bwd_weight_xdl_fp16 grouped_conv_bwd_weight_xdl_fp16.cpp)
+add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_fp16)
 
-        add_example_executable(example_grouped_conv_bwd_weight_xdl_bf16 grouped_conv_bwd_weight_xdl_bf16.cpp)
-        add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_bf16)
+add_example_executable(example_grouped_conv_bwd_weight_xdl_bf16 grouped_conv_bwd_weight_xdl_bf16.cpp)
+add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_bf16)
 
-        add_example_executable(example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8 grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8.cpp)
-        add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8)
-        set(target 1)
-    endif()
+add_example_executable(example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8 grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8.cpp)
+add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8)
 
-    if(gpu IN_LIST gpu_list_wmma AND target EQUAL 0)
-        add_custom_target(example_grouped_conv_bwd_weight)
-        add_example_executable(example_grouped_conv_bwd_weight_wmma_fp16 grouped_conv_bwd_weight_wmma_fp16.cpp)
-        add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_wmma_fp16)
-        set(target 1)
-    endif()
-endforeach()
-
-add_custom_target(example_grouped_conv_bwd_weight_dl)
+add_example_executable(example_grouped_conv_bwd_weight_wmma_fp16 grouped_conv_bwd_weight_wmma_fp16.cpp)
+add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_wmma_fp16)
 
 add_example_executable(example_grouped_conv_bwd_weight_dl_fp16 grouped_conv_bwd_weight_dl_fp16.cpp)
-add_example_dependencies(example_grouped_conv_bwd_weight_dl example_grouped_conv_bwd_weight_dl_fp16)
+add_example_dependencies(example_grouped_conv_bwd_weight example_grouped_conv_bwd_weight_dl_fp16)

example/21_gemm_layernorm/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-   add_example_executable(example_gemm_bias_relu_add_layernorm_xdl_welford_fp16 gemm_bias_relu_add_layernorm_xdl_welford_fp16.cpp)
-   add_example_executable(example_gemm_bias_relu_add_layernorm_xdl_naive_fp16 gemm_bias_relu_add_layernorm_xdl_naive_fp16.cpp)
-   add_example_executable(example_gemm_layernorm_xdl_naive_fp16 gemm_layernorm_xdl_naive_fp16.cpp)
-   add_example_executable(example_gemm_xdl_layernorm_naive_single_kernel_fp16 gemm_xdl_layernorm_naive_single_kernel_fp16.cpp)
-   set(target 1)
- endif()
-endforeach()
-
+add_example_executable(example_gemm_bias_relu_add_layernorm_xdl_welford_fp16 gemm_bias_relu_add_layernorm_xdl_welford_fp16.cpp)
+add_example_executable(example_gemm_bias_relu_add_layernorm_xdl_naive_fp16 gemm_bias_relu_add_layernorm_xdl_naive_fp16.cpp)
+add_example_executable(example_gemm_layernorm_xdl_naive_fp16 gemm_layernorm_xdl_naive_fp16.cpp)
+add_example_executable(example_gemm_xdl_layernorm_naive_single_kernel_fp16 gemm_xdl_layernorm_naive_single_kernel_fp16.cpp)

example/26_contraction/CMakeLists.txt

@@ -4,49 +4,49 @@ add_custom_target(example_contraction_bilinear)
 
 # FP32
 add_example_executable(example_contraction_bilinear_xdl_fp32 contraction_bilinear_xdl_fp32.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32)
 
 add_example_executable(example_contraction_scale_xdl_fp32 contraction_scale_xdl_fp32.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32)
 
 add_example_executable(example_contraction_bilinear_xdl_fp32_compute_bf16 contraction_bilinear_xdl_fp32_compute_bf16.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32_compute_bf16)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32_compute_bf16)
 
 add_example_executable(example_contraction_scale_xdl_fp32_compute_bf16 contraction_scale_xdl_fp32_compute_bf16.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32_compute_bf16)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32_compute_bf16)
 
 add_example_executable(example_contraction_bilinear_xdl_fp32_compute_fp16 contraction_bilinear_xdl_fp32_compute_fp16.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32_compute_fp16)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp32_compute_fp16)
 
 add_example_executable(example_contraction_scale_xdl_fp32_compute_fp16 contraction_scale_xdl_fp32_compute_fp16.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32_compute_fp16)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp32_compute_fp16)
 
 # FP64
 add_example_executable(example_contraction_bilinear_xdl_fp64 contraction_bilinear_xdl_fp64.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp64)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp64)
 
 add_example_executable(example_contraction_scale_xdl_fp64 contraction_scale_xdl_fp64.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp64)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp64)
 
 add_example_executable(example_contraction_bilinear_xdl_fp64_compute_fp32 contraction_bilinear_xdl_fp64_compute_fp32.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp64_compute_fp32)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp64_compute_fp32)
 
 add_example_executable(example_contraction_scale_xdl_fp64_compute_fp32 contraction_scale_xdl_fp64_compute_fp32.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp64_compute_fp32)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp64_compute_fp32)
 
 # FP16
 add_example_executable(example_contraction_bilinear_xdl_fp16_compute_fp32 contraction_bilinear_xdl_fp16_compute_fp32.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp16_compute_fp32)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_fp16_compute_fp32)
 
 add_example_executable(example_contraction_scale_xdl_fp16_compute_fp32 contraction_scale_xdl_fp16_compute_fp32.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_fp16_compute_fp32)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_fp16_compute_fp32)
 
 # BF16
 add_example_executable(example_contraction_bilinear_xdl_bf16_compute_fp32 contraction_bilinear_xdl_bf16_compute_fp32.cpp)
-add_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_bf16_compute_fp32)
+add_example_dependencies(example_contraction_bilinear example_contraction_bilinear_xdl_bf16_compute_fp32)
 
 add_example_executable(example_contraction_scale_xdl_bf16_compute_fp32 contraction_scale_xdl_bf16_compute_fp32.cpp)
-add_dependencies(example_contraction_scale example_contraction_scale_xdl_bf16_compute_fp32)
+add_example_dependencies(example_contraction_scale example_contraction_scale_xdl_bf16_compute_fp32)
 
-add_dependencies(example_contraction example_contraction_scale)
-add_dependencies(example_contraction example_contraction_bilinear)
+add_example_dependencies(example_contraction example_contraction_scale)
+add_example_dependencies(example_contraction example_contraction_bilinear)

example/29_batched_gemm_bias_e_permute/CMakeLists.txt

 add_example_executable(example_batched_gemm_bias_e_permute_xdl_fp16 batched_gemm_bias_e_permute_xdl_fp16.cpp)
-
-if(GPU_TARGETS MATCHES "gfx11")
-    add_example_executable(example_batched_gemm_bias_e_permute_wmma_fp16 batched_gemm_bias_e_permute_wmma_fp16.cpp)
-endif()
+add_example_executable(example_batched_gemm_bias_e_permute_wmma_fp16 batched_gemm_bias_e_permute_wmma_fp16.cpp)

example/30_grouped_conv_fwd_multiple_d/CMakeLists.txt

-list(APPEND gpu_list1 gfx908 gfx90a gfx940 gfx941 gfx942)
-list(APPEND gpu_list2 gfx1100 gfx1101 gfx1102)
+add_custom_target(example_grouped_conv_fwd_multiple_d)
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_fp16 grouped_conv_fwd_bias_relu_add_xdl_fp16.cpp)
+add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_fp16)
 
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list1 AND target EQUAL 0)
-        add_custom_target(example_grouped_conv_fwd_multiple_d)
+add_example_executable(example_grouped_conv_fwd_xdl_fp16 grouped_conv_fwd_xdl_fp16.cpp)
+add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_xdl_fp16)
 
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_fp16 grouped_conv_fwd_bias_relu_add_xdl_fp16.cpp)
-        add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_fp16)
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_fp32 grouped_conv_fwd_bias_relu_add_xdl_fp32.cpp)
+add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_fp32)
 
-        add_example_executable(example_grouped_conv_fwd_xdl_fp16 grouped_conv_fwd_xdl_fp16.cpp)
-        add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_xdl_fp16)
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_bf16 grouped_conv_fwd_bias_relu_add_xdl_bf16.cpp)
+add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_bf16)
 
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_fp32 grouped_conv_fwd_bias_relu_add_xdl_fp32.cpp)
-        add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_fp32)
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_int8 grouped_conv_fwd_bias_relu_add_xdl_int8.cpp)
+add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_int8)
 
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_bf16 grouped_conv_fwd_bias_relu_add_xdl_bf16.cpp)
-        add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_bf16)
+if(USE_BITINT_EXTENSION_INT4)
+    add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_int4 grouped_conv_fwd_bias_relu_add_xdl_int4.cpp)
+    add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_int4)
+endif() # USE_BITINT_EXTENSION_INT4
 
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_int8 grouped_conv_fwd_bias_relu_add_xdl_int8.cpp)
-        add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_int8)
-
-        if(USE_BITINT_EXTENSION_INT4)
-            add_example_executable(example_grouped_conv_fwd_bias_relu_add_xdl_int4 grouped_conv_fwd_bias_relu_add_xdl_int4.cpp)
-            add_example_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_int4)
-        endif() # USE_BITINT_EXTENSION_INT4
-
-        set(target 1)
-    endif()
-endforeach()
-
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
-    if(gpu IN_LIST gpu_list2 AND target EQUAL 0)
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_wmma_fp16 grouped_conv_fwd_bias_relu_add_wmma_fp16.cpp)
-        add_example_executable(example_grouped_conv_fwd_bias_relu_add_wmma_int8 grouped_conv_fwd_bias_relu_add_wmma_int8.cpp)
-        set(target 1)
-    endif()
-endforeach()
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_wmma_fp16 grouped_conv_fwd_bias_relu_add_wmma_fp16.cpp)
+add_example_executable(example_grouped_conv_fwd_bias_relu_add_wmma_int8 grouped_conv_fwd_bias_relu_add_wmma_int8.cpp)

example/31_batched_gemm_gemm/CMakeLists.txt

-list(APPEND gpu_list1 gfx908 gfx90a gfx940 gfx941 gfx942)
-
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list1 AND target EQUAL 0)
-   add_example_executable(example_batched_gemm_gemm_xdl_fp32 batched_gemm_gemm_xdl_fp32.cpp)
-   add_example_executable(example_batched_gemm_gemm_xdl_fp16 batched_gemm_gemm_xdl_fp16.cpp)
-   add_example_executable(example_batched_gemm_gemm_xdl_bf16 batched_gemm_gemm_xdl_bf16.cpp)
-   if(USE_BITINT_EXTENSION_INT4)
-      add_example_executable(example_batched_gemm_gemm_xdl_int4 batched_gemm_gemm_xdl_int4.cpp)
-   endif(USE_BITINT_EXTENSION_INT4)
-   set(target 1)
- endif()
-endforeach()
+add_example_executable(example_batched_gemm_gemm_xdl_fp32 batched_gemm_gemm_xdl_fp32.cpp)
+add_example_executable(example_batched_gemm_gemm_xdl_fp16 batched_gemm_gemm_xdl_fp16.cpp)
+add_example_executable(example_batched_gemm_gemm_xdl_bf16 batched_gemm_gemm_xdl_bf16.cpp)
+if(USE_BITINT_EXTENSION_INT4)
+   add_example_executable(example_batched_gemm_gemm_xdl_int4 batched_gemm_gemm_xdl_int4.cpp)
+endif(USE_BITINT_EXTENSION_INT4)
 
 if(NOT GPU_TARGETS MATCHES "gfx94" AND NOT GPU_TARGETS MATCHES "gfx1")
    add_example_executable(example_batched_gemm_gemm_xdl_int8 batched_gemm_gemm_xdl_int8.cpp)

example/32_batched_gemm_scale_softmax_gemm/CMakeLists.txt

-if(GPU_TARGETS MATCHES "gfx11")
-    add_example_executable(example_batched_gemm_lower_triangle_scale_softmax_gemm_permute_wmma_fp16 batched_gemm_lower_triangle_scale_softmax_gemm_permute_wmma_fp16.cpp)
-    add_example_executable(example_batched_gemm_scale_softmax_gemm_permute_wmma_fp16 batched_gemm_scale_softmax_gemm_permute_wmma_fp16.cpp)
-    add_example_executable(example_self_attention_forward_wmma_fp16 self_attention_forward_wmma_fp16.cpp)
-    add_example_executable(example_cross_attention_forward_wmma_fp16 cross_attention_forward_wmma_fp16.cpp)
-    add_example_executable(example_multi_query_attention_forward_wmma_fp16 multi_query_attention_forward_wmma_fp16.cpp)
-    add_example_executable(example_grouped_query_attention_forward_wmma_fp16 grouped_query_attention_forward_wmma_fp16.cpp)
-endif()
+add_example_executable(example_batched_gemm_lower_triangle_scale_softmax_gemm_permute_wmma_fp16 batched_gemm_lower_triangle_scale_softmax_gemm_permute_wmma_fp16.cpp)
+add_example_executable(example_batched_gemm_scale_softmax_gemm_permute_wmma_fp16 batched_gemm_scale_softmax_gemm_permute_wmma_fp16.cpp)
+add_example_executable(example_self_attention_forward_wmma_fp16 self_attention_forward_wmma_fp16.cpp)
+add_example_executable(example_cross_attention_forward_wmma_fp16 cross_attention_forward_wmma_fp16.cpp)
+add_example_executable(example_multi_query_attention_forward_wmma_fp16 multi_query_attention_forward_wmma_fp16.cpp)
+add_example_executable(example_grouped_query_attention_forward_wmma_fp16 grouped_query_attention_forward_wmma_fp16.cpp)
 
 add_custom_target(example_gemm_scale_softmax_gemm)
 

example/35_splitK_gemm/CMakeLists.txt

-list(APPEND gpu_list gfx908 gfx90a gfx940 gfx941 gfx942)
-set(target 0)
-foreach(gpu IN LISTS GPU_TARGETS)
- if(gpu IN_LIST gpu_list AND target EQUAL 0)
-   add_custom_target(example_splitK_gemm_xdl)
+add_custom_target(example_splitK_gemm_xdl)
+add_example_executable(example_splitK_gemm_xdl_fp32 splitK_gemm_xdl_fp32.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp32)
 
-   add_example_executable(example_splitK_gemm_xdl_fp32 splitK_gemm_xdl_fp32.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp32)
+add_example_executable(example_splitK_gemm_xdl_fp16 splitK_gemm_xdl_fp16.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16)
 
-   add_example_executable(example_splitK_gemm_xdl_fp16 splitK_gemm_xdl_fp16.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16)
+add_example_executable(example_splitK_gemm_xdl_fp16_fp8 splitK_gemm_xdl_fp16_fp8.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16_fp8)
 
-   add_example_executable(example_splitK_gemm_xdl_fp16_fp8 splitK_gemm_xdl_fp16_fp8.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_fp16_fp8)
+add_example_executable(example_splitK_gemm_xdl_lds_direct_load_fp16 splitK_gemm_xdl_lds_direct_load_fp16.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_lds_direct_load_fp16)
 
-   add_example_executable(example_splitK_gemm_xdl_lds_direct_load_fp16 splitK_gemm_xdl_lds_direct_load_fp16.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_lds_direct_load_fp16)
+add_example_executable(example_splitK_gemm_xdl_bf16 splitK_gemm_xdl_bf16.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_bf16)
 
-   add_example_executable(example_splitK_gemm_xdl_bf16 splitK_gemm_xdl_bf16.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_bf16)
+add_example_executable(example_splitK_gemm_xdl_int8 splitK_gemm_xdl_int8.cpp)
+add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int8)
 
-   add_example_executable(example_splitK_gemm_xdl_int8 splitK_gemm_xdl_int8.cpp)
-   add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int8)
-
-   if(USE_BITINT_EXTENSION_INT4)
-       add_example_executable(example_splitK_gemm_xdl_int4 splitK_gemm_xdl_int4.cpp)
-       add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int4)
-   endif()
-
-   set(target 1)
- endif()
-endforeach()
+if(USE_BITINT_EXTENSION_INT4)
+    add_example_executable(example_splitK_gemm_xdl_int4 splitK_gemm_xdl_int4.cpp)
+    add_example_dependencies(example_splitK_gemm_xdl example_splitK_gemm_xdl_int4)
+endif()